My personal favourite for taking out such information in Photoshop is to use the Patch tool. Copying over a piece of the background over the letters/numbers erases them in such a way that looks like they never existed before.
They may have had their Cultural Revolution that happened back in the 60's and 70's, but they still do not grasp the base logic of having a strong nation;
Well, DUH! The Cultural Revolution was an unmitigated disaster, setting China back decades culturally, economically, and intellectually. And this isn't just a view from an outsider, the current Chinese government considers it bad as well. Education can to a standstill for TEN years, making an entire GENERATION of Chinese inadequately educated. Intellectuals during the period were systematically killed. There were reports of people beaten to death by mobs and then eaten afterwards.
If you read the cygwin mailing list archives (http://sourceware.org/ml/cygwin/2002-04/msg01071.html) it looks like you can on Windows NT/XP. However, it doesn't sound like they went through the trouble to figure out why their implementation was actually slower because it wasn't compatible with Win95.
Your problem is that you never *asked* in the first place. I *always* ask first what are the trade-offs (eg. performance, reliability, whatever else that may seem relevant to the problem) that they want to see in my answer. In the real world, users never really tell you what they want (and often they don't know themselves). IMHO, it's part of the job. By asking first, it shows that you think first, before acting.
The nice thing to do would be to tell them you won't hire them, exactly why you won't hire them, and tell them what they would need to work on so you would hire them next year. The person metioned in GP managed to send in a resume with 4(!) spelling mistakes on their resume. This means attention to detail is clearly lacking. Fixing that up on their resume won't make me want to hire them next year, or ever. This is not like some post that he typed up on slashdot, this is his *resume*. If the person can't get that right before sending it in, there's no point.
That was the worst code I've seen outside of a classroom. The writer, for some bizarre reason, had been promoted to project manager in spite of the fact that he had only been a software developer for a couple of years or so. If he's that bad at coding, I'd promote him to project manager just to keep him from coding!
A more plausible claim, and one much harder to prove or disprove, is that the Office team has access to Windows source code, so that rather than creating something from scratch they can just grab a copy of the menuing code and create their own version. I always assumed this was true. Why would they not?
Re:Mentioning that you were involved with VRML...
on
Collada
·
· Score: 1
Who, exactly, do they think is going to use this besides amateurs and little tools companies (like the ones linked to in the article) who cater to amateurs...?
I know of at least two production companies that use a proprietary XML format. ( http://www.xsi-blog.com/?p=15 ). If Collada had existed at the time, maybe they would have used Collada instead.
> Because a key requirement was commercial support == you call someone to fix bugs for you, not fix > them your self. I think the main problem was not shopping around for the required support.
Did you read the GP? Go buy a $140 copy of WinXP and tell me how it goes with calling Microsoft for *any* type of support, much less so for even trying to fix bugs for you.
One of the reasons Microsoft moved a lot of functionality into the Kernel between NT 3.5 and NT4.0 was performances (NT being, at its origins a uK based OS).
And I curse them to this day for doing it! NT 3.51 was rock-solid. NT 4 was a piece of crap that crashed all the time. And if you've been following the Vista stuff, they've started moving parts back out of the kernel. From the horse's mouth: http://channel9.msdn.com/ShowPost.aspx?PostID=1488 20
If you're on Windows, the first line of defense is running a debug build of the app with/RTC1 and/GZ turned on. I ran into a stack buffer overrun just the other day and message box came up immediately telling me exactly which variable got corrupted.
Re:Effective cache use will be a better optimisati
on
Arrays vs Pointers in C?
·
· Score: 2, Funny
I have good story for this one. In my second co-op term, I was asked to improve the speed of the general 2D convolution image filter for a well-known commercial paint package. Nothing complicated, just run the given 2D kernel through the image.
The previous writer of the function had carefully removed all the array accesses, only using pointers that were incremented by fixed constants as it proceeded through the code. He had also carefully maintained an array of the sums of the results as we moved from pixel to pixel to avoid re-calculating anything twice. You would think, it's pretty fast, eh?
I came across two problems with the code:
1. His array of sums was actually a queue and so he shifted the entire array by one element for every single pixel. Using a ring buffer instead quickly solved this one.
2. I then came to realize that he was traversing the row-major image in columns. The cache coherency was being shot to hell because none of the cache-lines were being hit as it went from pixel to pixel. I rewrote the function to go by rows instead and guess what the speed improvement was? Something like 3 times!
Go figure. Some people optimize and get it entirely wrong.
Interesting. Which "shell calls" are we talking about exactly? It would be nice to know which these are and then spawn a thread to handle them if possible.
> in theory, you can run any imaginable query, > but in reality, you can't touch an un-indexed > field on most production databases unless you've > got *lots* of horsepower to burn, and very > patient (read: non-existent) users.
If you've got more than 10's of gigabytes of raw data to search, and are willing to trade indexing time for query, these products can index ALL your data without exponential increase while achieving order of magnitude faster queries: http://www.arunasoftware.com/ http://ww w.disc.com/products.html
I don't this. Why didn't Linus use Perforce then? It's been free for opensource use since conception and it seems to beats all the other commercial scms.
Slashdot Mirror
My personal favourite for taking out such information in Photoshop is to use the Patch tool. Copying over a piece of the background over the letters/numbers erases them in such a way that looks like they never existed before.
They may have had their Cultural Revolution that happened back in the 60's and 70's, but they still do not grasp the base logic of having a strong nation;
Well, DUH! The Cultural Revolution was an unmitigated disaster, setting China back decades culturally, economically, and intellectually. And this isn't just a view from an outsider, the current Chinese government considers it bad as well. Education can to a standstill for TEN years, making an entire GENERATION of Chinese inadequately educated. Intellectuals during the period were systematically killed. There were reports of people beaten to death by mobs and then eaten afterwards.
If you read the cygwin mailing list archives (http://sourceware.org/ml/cygwin/2002-04/msg01071.html) it looks like you can on Windows NT/XP. However, it doesn't sound like they went through the trouble to figure out why their implementation was actually slower because it wasn't compatible with Win95.
I think the GP meant what if the program generating the schedules was compromised? Then it could generate a "random" schedule that the attacker knows.
Your problem is that you never *asked* in the first place. I *always* ask first what are the trade-offs (eg. performance, reliability, whatever else that may seem relevant to the problem) that they want to see in my answer. In the real world, users never really tell you what they want (and often they don't know themselves). IMHO, it's part of the job. By asking first, it shows that you think first, before acting.
It doesn't have to UI code, which is besides the point.
Who, exactly, do they think is going to use this besides amateurs and little tools companies (like the ones linked to in the article) who cater to amateurs...?
I know of at least two production companies that use a proprietary XML format. ( http://www.xsi-blog.com/?p=15 ). If Collada had existed at the time, maybe they would have used Collada instead.
Excellent, I hope they sue Microsoft! :)
Try also the Handbook of Algorithms and Data Structures:
r uctures-Pascal/dp/0201416077/sr=8-1/qid=1160102805 /ref=sr_1_1/102-2768899-0462565?ie=UTF8&s=books
http://www.amazon.com/Handbook-Algorithms-Data-St
> Because a key requirement was commercial support == you call someone to fix bugs for you, not fix
> them your self. I think the main problem was not shopping around for the required support.
Did you read the GP? Go buy a $140 copy of WinXP and tell me how it goes with calling Microsoft for *any* type of support, much less so for even trying to fix bugs for you.
One of the reasons Microsoft moved a lot of functionality into the Kernel between NT 3.5 and NT4.0 was performances (NT being, at its origins a uK based OS).
8 20
And I curse them to this day for doing it! NT 3.51 was rock-solid. NT 4 was a piece of crap that crashed all the time. And if you've been following the Vista stuff, they've started moving parts back out of the kernel. From the horse's mouth: http://channel9.msdn.com/ShowPost.aspx?PostID=148
Mod parent up. I've used QuantLib and it's so cool!
You should check out this sometime:
http://www.arunasoftware.com/
If you're on Windows, the first line of defense is running a debug build of the app with /RTC1 and /GZ turned on. I ran into a stack buffer overrun just the other day and message box came up immediately telling me exactly which variable got corrupted.
I have good story for this one. In my second co-op term, I was asked to improve the speed of the general 2D convolution image filter for a well-known commercial paint package. Nothing complicated, just run the given 2D kernel through the image.
The previous writer of the function had carefully removed all the array accesses, only using pointers that were incremented by fixed constants as it proceeded through the code. He had also carefully maintained an array of the sums of the results as we moved from pixel to pixel to avoid re-calculating anything twice. You would think, it's pretty fast, eh?
I came across two problems with the code:
1. His array of sums was actually a queue and so he shifted the entire array by one element for every single pixel. Using a ring buffer instead quickly solved this one.
2. I then came to realize that he was traversing the row-major image in columns. The cache coherency was being shot to hell because none of the cache-lines were being hit as it went from pixel to pixel. I rewrote the function to go by rows instead and guess what the speed improvement was? Something like 3 times!
Go figure. Some people optimize and get it entirely wrong.
I'd like to chime in support of this. Having gone down the exact road of implementing something in UDP, make very sure that you absolutely need it.
Are you telling me that *ALL* Win32 API calls are handled through explorer.exe!? That's ridiculous.
Interesting. Which "shell calls" are we talking about exactly? It would be nice to know which these are and then spawn a thread to handle them if possible.
What's the biggest project you've ever worked on? Ever tried to rollback a sweep across 20 directories in CVS? God help you.
> in theory, you can run any imaginable query,
w w.disc.com/products.html
> but in reality, you can't touch an un-indexed
> field on most production databases unless you've
> got *lots* of horsepower to burn, and very
> patient (read: non-existent) users.
If you've got more than 10's of gigabytes of raw data to search, and are willing to trade indexing time for query, these products can index ALL your data without exponential increase while achieving order of magnitude faster queries:
http://www.arunasoftware.com/
http://w
I don't this. Why didn't Linus use Perforce then? It's been free for opensource use since conception and it seems to beats all the other commercial scms.
Perforce also uses this concept of a "transaction" which I think predates BitKeeper?